This letter deals with precision issues in the determination of the timing of the room impulse responses (RIRs) onset. First, it is shown that while errors of onset timing estimation do not have that much effect on temporal indices, an erroneous onset estimation leads to significant differences in energetic and statistic acoustical indices. Twelve automatic onset detection methods are compared, in terms of precision, robustness, and complexity. Experimental validation made on a set of 100 RIRs provides good evidence in favor of spectral and/or energetic methods, according to the type of sound source.